Enhancement of RPC Positioning Accuracy Using Affine Transformation with Different Number of Ground Control Points

Document Type : Original Article

Authors

1 Egyptian Armed Forces, Egypt.

2 Civil Engineering Department, Faculty of Engineering, Ain Shams University, Cairo, Egypt.

Abstract

The spatial data of satellite image is determined by integration of geometry of imaging sensor and positioning sensors as GPS and star tracking systems. Sensor model is obtained using image geometry and system calibration parameters. Sensor model is not available but Rational Polynomial Function (RPC) model is provided as an alternative representation of sensor model. RPC model is used in determining spatial data for all features in the captured satellite image. The accuracy of geometric corrected image using RPC model may not be accepted for different applications. The availability of Ground Control Points (GCPs) is a motivation of this research paper in enhancing the accuracy of resultant RPC geometric correction image using GCPs. The available data is a high resolution satellite image which is captured by IKONOS2 senor. The RPC file with its parameters is available with seven GCPs from ground survey with high accuracy. The RPC model is used to get positioning information of the raw image pixels. The accuracy of the resultant image is tested against available GCPs. Some of available GCPs are added to RPC model with 1st order polynomial mathematical model (affine transformation) in image geometric correction process using ERDAS imagine software. The accuracy for each case of the resultant corrected image using both RPC and GCPs is tested against all the seven available GCPs. The accuracy assessment of the accuracy of resultant image is applied to get not only the improvement quality but also the suitable number of GCPs for enhancing accuracy of RPC geometric correction. Adding GCPs enhanced the accuracy of RPC geometric correction from around (10.0:20.0) meter accuracy to (1.0:5.0) meter accuracy with 80% improvement factor.

Keywords